“That’s funny,” Alexander Fleming famously remarked, upon noticing the bacteria-free ring around mould contamination in a Petri dish he’d left out. His accidental discovery of penicillin saved millions of lives. The next part of the story is often skipped; Fleming tried for over a decade to do anything useful with penicillin, and failed. He was unable to cultivate the mould in the necessary quantities, isolate the substance and identify an appropriate method of delivery. Fleming’s experiments suggested that penicillin would leave the body too quickly to be useful, and early trials of penicillin applied directly to wounds were failures. Scientists at the Dunn School eventually succeeded in producing small amounts and settled on the method of injection, but watched their first patient die; though penicillin produced an initial improvement, the lab could not manufacture enough penicillin to save him despite employing every available bathtub and bedpan and even recycling his urine.
Pharmaceutical development has never been an easy process.
Of course, some medicines are relatively easy to produce. Poppy seeds have been used as an effective painkiller since the days of Ancient Egypt; nowadays we know their contents as morphine and codeine. Anyone can grow tea leaves or wine grapes. But even when drugs are found in the natural world around us, they rarely come ready-made. The elegant purple flowering plants which contain an antidote to sarin gas, also known as deadly nightshade, aren’t exactly safe to just pick and eat.
The typical cost of bringing a drug to market, according to researchers at the Tufts Centre for the Study of Drug Development, is $2.6 billion, $1.4 billion of which is out-of-pocket expense. Costs are so high because for every drug that receives approval, hundreds or thousands more are tried and written off. Costs have also been increasing with rising safety standards, increasing researcher ambition and a growing attitude that scientific problems are best solved by burying them in cash.
Luckily, modern pharmacology doesn’t have to rely on leaving Petri dishes out and hoping for the best. Increasingly, pharmaceutical companies are using powerful computers and rapidly advancing biological knowledge to conduct “structure-based drug design”. This strategy focuses on precise models of the mechanisms by which a disease causes its negative effects on the body, right down at the molecular level. Genes, proteins and enzymes from healthy and sick cells are compared so scientists can identify methods of interfering with unhealthy processes or restoring normal signalling pathways.
Unfortunately for the curious non-expert, biological pathways often involve incredibly complex cascades of chemical reactions facilitated by enzymes with names longer than your arm. Take insulin, the sugar-regulating hormone made famous by diabetes. In healthy bodies, glucose in the blood causes certain pancreatic cells to release insulin, which binds to receptors (parts of a cell which can cause reactions inside the cell when they detect something outside). When this happens the receptors cause cells to produce other chemicals called Ptdlns (3,4,5)P3 which indirectly activate “protein kinase B”, which then interacts with various other chemicals in the liquid and the nucleus of the cell. This leads to the deactivation of GSK3, a chemical which prevents the action of an enzyme (a biological catalyst) called glycogen synthase. When GSK3 is deactivated, glycogen synthase is able to catalyse the conversion of glucose to glycogen – the body’s way of putting sugar into long-term storage. And this is just one of the numerous effects of this process; insulin also causes changes in how muscles absorb sugar, how genes are expressed and how proteins are made. Our bodies put even the craziest Rube Goldberg machine to shame.
Pharmacologists might study any of the chemicals involved in biological processes like that one, with cutting-edge sensors and software which can allow them to understand the structure of the molecules. Targets are often enzymes, the catalysts which help fit other molecules together, or receptors, the proteins in cell membranes which allow signals outside the cell to cause responses inside it. Once a potential target (whether an enzyme, a receptor or a protein) has been studied, the task is to find a chemical (or “ligand”) which could bind to it. The goal might be to find an ‘agonist’, which causes a reaction when it binds to the receptor, or an ‘antagonist’ (also known as a ‘blocker’), which occupies the receptor so that agonists can’t do their work. It’s like designing a key to fit a lock – except the lock is tiny and shaped like a horrifically tangled ball of string.
The ‘key’ needs to not only fit the lock, but also not fit too many other locks. This quality is called “specificity” and it is important in reducing the seriousness of side effects. The search for specificity is a significant part of the difficulty in tackling cancer; since cancer cells are identical to the body’s own cells in many ways, it’s difficult to kill them without killing the patient (generally considered a fairly unacceptable side effect).
Specificity isn’t only a problem for cancer drugs. The issue is that many hormones are involved in a startling variety of biological functions, thanks to the tack-something-extra-on-top nature of evolution’s innovations. As an example, phospholipases free arachidonic acid from fats in cell membranes, which allows cyclooxygenase (COX) enzymes to process the acid into prostaglandins. Ibuprofen and similar drugs work by competing with the arachidonic acid, binding to cyclooxygenase temporarily occupying the ‘active site’ and inhibiting the production of prostaglandins. Prostaglandins interact with receptors in cells and cause pain signals, arterial dilation and blood clotting – so preventing their production also prevents pain and inflammation. Medicines which either increase or decrease prostaglandins are used for arthritis, injuries, labour, heavy menstruation, cancer, stomach ulcers and glaucoma. Given the sheer variety of their effects, it isn’t surprising that ibuprofen has side effects on the digestive system – prostaglandins are involved in regulating blood flow through the gastrointestinal tract.
Advanced imaging techniques might be used to test 5,000 to 10,000 compounds in order to find just one candidate medicine. Many drugs will fail before ever being tested on humans, if they are shown to harm mice or cause adverse reactions to human cells grown in test tubes.
When a drug seems to work in the laboratory, clinical research – tests on humans – can begin. The regulated structure of these tests is the result of a trade-off between the moral imperative to expose as few people as possible to a potentially unsafe drug, the scientific requirement to use a large sample size, and the financial impossibility of funding infinite research.
Phase I trials are done on small (20-100) groups of people. Usually the drug is given to healthy people, with the only goal being to supervise them closely and show it is safe enough to test on sick people. Approximately 63% of drugs which reach Phase I move on to Phase II, which is the stage where experimental medicine can be given to sick people at hopefully-effective doses. The tests only involve a few hundred subjects – enough to suggest a drug works, but nowhere near enough to prove it.
Only 31% of Phase II trials are successful. The others show that the drug in question is ineffective, that it has safety issues, or that the side effects aren’t worth the cure. It costs $7-20 million to run a Phase II trial – clinic space must be hired, volunteers recruited and expensed, doctors paid, drugs manufactured, medical equipment bought and mountains of paperwork scaled – which often raises questions about how to justify the investment given the likelihood of failure.
Phase III is the point at which drugs can, finally, actually be tested. Thousands of volunteers, often across multiple international sites, make up a sample large enough to provide statistical evidence of a drug’s effects. 58% of Phase III trials gain enough data on the safety and efficacy to make an application for licensing, of which 85% eventually receive approval.11
It still isn’t clear how the licensing process will work in the UK after Brexit; most applications are made to the USA’s Food and Drug Administration and the European Medicines Agency. The Medicines and Healthcare products Regulatory Agency (MHRA) will become solely responsible for licensing medicines in the UK, and intends to grandfather in existing EMA approvals and temporarily continue following some EU regulations, such as approving CE-marked medical devices and using EU Good Practice requirements. Nobody knows exactly how the MHRA might change the rules, or whether global companies will consider MHRA approval attractive enough to prioritise.
Out of 100 products that enter Phase I trials, approximately 63 will move on to Phase II, approximately 20 to Phase III, will make a marketing application and 10 will receive approval. The overall 9.7% success rate reported by BIO may seem low, but it is impressive considering the near-zero rate at which randomly chosen substances from your garden or kitchen would cure rather than cause an upset stomach, reflecting the huge numbers of candidates rejected before Phase I. The rate also disproportionately reflects the poor outlook for cancer and neurology, focus areas which attract many trials and produce many failures; medicines for hematology have a success rate of 26% from Phase I to approval. As it turns out, curing cancer is hard.
Still, the cost and difficulty of drug development is reflected in the number of products available. Only 35 new active substances were authorized by the European Medicines Agency in 2017. 6 were denied, 14 applications were voluntarily withdrawn, and nobody knows how many were deterred from applying in the first place.
Almost a third of the approved medicines were for cancer, and only four were for infectious disease. Thirteen substances were approved through the Orphan Medicine Program, which selectively offers support to the development of drugs which treat rare diseases. These programs, and the need to generate profits to cover development costs from selling to the wealthy, provide clear incentives for the pharmaceutical industry to focus on rare non-communicable diseases which primarily afflict the elderly. Relatively little drug development is targeted at diseases like malaria or diarrheal diseases, despite being among the leading killers of children worldwide and generally easier to treat than cancer.
After approval, clinical trials don’t stop. Some safety issues might not be obvious until the drug is given to far more people than is viable in a clinical trial – for instance, if 1 in 10,000 people has an allergic reaction to the drug, this won’t be discovered until Phase IV (also known as ‘post-marketing surveillance’). However, the drug can be marketed and sold to a wide range of people while this research is ongoing.
This is the point at which pharmaceutical companies begin racing to generate enough profit to cover their costs before the patent expires. Only one in five approved drugs will succeed at this task. As many as 25,000 compounds may start in the laboratory for every one that recoups what was invested.
Of course, pharmaceutical development has never been an easy process. Perhaps the only thing even more impressive than the sheer number of hurdles to developing drugs is science’s ability to overcome those hurdles. The first clinical trial was not performed until 1747, and it was an unimpressive affair – James Lind administered fruit, among various other substances, to sailors suffering from scurvy in the miserable conditions of a long sea voyage. The lab kits, advanced statistics and biomarker tests used today have played a vital part in ending harmful treatments from bloodletting to prescribing thalidomide for morning sickness. Fleming, whose first work was in showing that surface antiseptics killed more people than they saved, would have been proud.